Maternal and Child Health Inequalities in Ethiopia

Policy Research Working Paper 7508

Maternal and Child Health Inequalities in Ethiopia

Alemayehu AmbelColin AndrewsAnne BakilanaElizabeth Foster

Qaiser KhanHuihui Wang

Social Protection and Labor Global Practice GroupDecember 2015

WPS7508

Produced by the Research Support Team

Abstract

The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent.

Policy Research Working Paper 7508

This paper is a product of the Social Protection and Labor Global Practice Group. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org. The authors may be contacted at [email protected].

Recent surveys show considerable progress in maternal and child health in Ethiopia. The improvement has been in health outcomes and health services coverage. The study examines how different groups have fared in this progress. It tracked 11 health outcome indicators and health inter-ventions related to Millennium Development Goals 1, 4, and 5. These are stunting, underweight, wasting, neonatal mortality, infant mortality, under-five mortality, measles vaccination, full immunization, modern contraceptive use by currently married women, antenatal care visits, and skilled birth attendance. The study explores trends in inequalities by household wealth status, mothers’ education,

and place of residence. It is based on four Demographic and Health Surveys implemented in 2000, 2005, 2011, and 2014. Trends in rate differences and rate ratios are analyzed. The study also investigates the dynamics of inequalities, using concentration curves for different years. In addi-tion, a decomposition analysis is conducted to identify the role of proximate determinants. The study finds substan-tial improvements in health outcomes and health services. Although there still exists a considerable gap between the rich and the poor, the study finds some reductions in inequalities of health services. However, some of the improve-ments in selected health outcomes appear to be pro-rich.

Maternal and Child Health Inequalities in Ethiopia

Alemayehu Ambel*, Colin Andrews, Anne Bakilana, Elizabeth Foster, Qaiser Khan and

Huihui Wang

Keywords: maternal and child health, health inequality, Ethiopia JEL classification: I14

* Alemayehu Ambel (Corresponding Author Email: [email protected]), Development Research Group, World Bank; Anne Bakilana and Huihui Wang, GHNDR, World Bank; Colin Andrews, Elizabeth Foster and Qaiser Khan, GSPDR, World Bank. The authors acknowledge financial support from the World Bank. The authors would like to thank: Phyllis Ronek for editorial assistance; Tekabe Belay, GNV Ramana and Getahun Tafesse (World Bank) for valuable comments and suggestions.

2

1. Introduction

Improving maternal and child health is integral to the Millennium Development Goals

(MDGs). In particular, Goals 4 and 5 call for, respectively, reduction in child and

maternal mortality for the period between 1990 and 2015. Goal 4 targeted a reduction

of under-five mortality by two-thirds, while Goal 5 set a 75 percent reduction in

maternal mortality. The goals received national and global attention over the last

decade and a half; countries and their international development partners mobilized

support to expand childhood immunization and increase availability and utilization of

maternal health services.

By the target date of 2015, different countries are at different stages in terms of

achieving the MDG targets. Country level tracking of the MDGs shows that Ethiopia is

one of those with satisfactory progress towards achieving the targets. A recent UN

report shows that all the MDGs were either on track or likely to be on track.1 A similar

tracking by the Center for Global Development (CGDEV) reported satisfactory progress

on all goals. The CGDEV tracking in 2011 ranks Ethiopia 33 of 137 countries, with an

MDG progress index of 4.5 (on a scale of zero to 8 points).2 The country’s progress in

the health sector is also documented in other recent surveys. For example, the 2014

mini-Demographic and Health Survey (DHS) shows reductions in child undernutrition

1 A UN tracking for Ethiopia shows that the country is on track on goals 1 and 4 and likely to be on track on goal 5.http://www.et.one.un.org/index.php?option=com_content&view=article&id=14&Itemid=488 2 MDG Progress Index: Gauging Country-Level Achievements, Center for Global Development (CGDEV) http://www.cgdev.org/page/mdg-progress-index-gauging-country-level-achievements accessed on 1/31/15.

3

and child mortality, as well as increases in coverage of maternal and child health

services including antenatal cares, contraceptive prevalence and skilled birth attendance

(CSA, 2014).3

Now with the MDG period coming to an end and new targets set in the

Sustainable Development Goals, studies are investigating not only if the countries

progressed towards achieving the targets, but also if the achievements were inclusive

(Wagstaff et al., 2014). Indeed, health inequalities matter for several economic and non-

economic reasons, as supported by vast empirical literature in this area. However, the

available evidence on health inequality studies is mixed and context specific; there have

been appreciable and widening inequalities in some countries and improvements in

others (e.g., Barros et al. 2012; Wagstaff et al., 2014). These studies also point out

important heterogeneities in the results arising from the type of health service or the

way it is delivered. For example, Barros et al. (2012) find that those services that are

traditionally delivered at health facilities, such as skilled birth attendant and antenatal

care visits, have more socioeconomic inequalities than those delivered by outreach or

mass campaigns. This would point to the role of various constraints that

disproportionately affect health care utilization by the poor. Similarly, a recent multi-

country study of progress in health MDGs by Wagstaff et al. (2014) finds that the

poor are more likely to face health risks, including child undernutrition and mortality,

3 According to the 2014 mini DHS Survey, the reductions in undernutrition and child mortality levels from 2000 and 2005 are considerable. Similarly, during the same period, there has been considerable expansion in maternal health service. However, trend should not be confused with the overall health status.

4

and less likely to receive key health services (p. 137). The study also notes disconnect

between health outcomes that were pro-rich and interventions that tended to be pro-

poor. Another study by Victora et al. (2012) analyzed time trends in inequalities of

selected health indicators from the MDGs.4 In general, they note prevalent pro-rich

inequalities, but services tend to be pro-poor when there are rapid changes in overall

coverage.

There are a number of other health inequality studies on low-income countries.

Some of these studies looked at one or more health status indicators (e.g., McKinnon et

al. 2014 and Quentin et al. 2014), while others focused on health intervention indicators

(e.g., Kruk et al. 2008). McKinnon et al. (2014) investigated wealth-related and

educational inequalities in neonatal mortality (NMR) for 24 low- and middle-income

countries with substantial heterogeneity in both the size and direction of NMR

inequalities between countries. They find that inequalities declined in most of the

studied countries. They also find pro-rich inequalities in NMR increased in a few

countries, including Ethiopia.

While the evidence in multi-country studies offers some useful insight, an in-

depth analysis of country-level inequalities would provide a more detailed description

and a focused message based on the factors that are particularly relevant to the

country’s context. There are a few relevant studies on health inequalities in Ethiopia.

Most of them analyzed one or two indicators at the national level (e.g., Onarheim et al.,

4 The indicators include skilled birth attendants, measles vaccination, and a composite coverage index, and examined coverage of a newly introduced intervention, use of insecticide-treated bednets by children. They use DHS and MICS data from 35 countries.

5

2013; World Bank, 2012; Yesuf and Calderon-Margalit, 2013), while others examined

one or two indicators for a specific region or city (e.g., Mirkuzie, 2014; Yesuf et al.,

2014). In this regard, a World Bank (2012) fact sheet on health equity and financial

protection on Ethiopia provides progress on a comprehensive list of health and health-

related indicators using multiple surveys from different sources.

In this study, we provide evidence on maternal and child health inequalities in

Ethiopia. We expand the available empirical evidence on maternal and child health

equity in Ethiopia by adding a more detailed inequality analysis using data from a

series of comparable surveys from 2000 to 2014. The surveys allow us to analyze the

trends in inequalities over a time period that overlaps with most of the MDGs’ years.

We also examine the contribution of the underlying socioeconomic determinants of

maternal and child health outcomes, such as wealth, place of residence and mothers’

education level.

The rest of the paper is organized as follows. Section 2 presents the methodology,

focusing on the indicators and their definition, statistical analysis and data. Section 3

presents the results. Section 4 discusses the results and provides a conclusion.

2. Methods

2.1 Indicators and definitions

Table 1 presents the health status and intervention indicators analyzed in this study.

Following recent health equity studies, we identified the indicators based on their

relevance to the health MDGs and availability of data (Barros et al., 2012; Victora et al.,

6

2012; Wagstaff et al., 2014). We focused on those MDG indicators that are relevant to

maternal and child health. These include Goal 4 (reducing child mortality) and Goal 5

(improving maternal mortality). Goal 1 is relevant as well, owing to its role in reduction

of undernutrition and hunger.

Table 1. List of maternal and child health indicators analyzed in this study

Indicators Definition

Stunting Percentage of children with a height-for-age z-score <-2 standard deviations from the reference median *

Wasting Percentage of children with a weight-for-height z-score <-2 standard deviations from the reference median *

Underweight Percentage of children with a weight-for-age z-score <-2 standard deviations from the reference median *

Neonatal mortality rate (NMR) The number of neonates dying before reaching 28 days of age per 1000 live births

Infant mortality rate (IMR) The number of deaths among children under 12 months of age per 1000 live births

Under-five mortality rate (U5MR) The number of deaths among children under 5 years of age per 1000 live births

Full immunization Percentage of children aged 12 to 23 months who received BCG, measles and three doses of Polio and DPT #,+

Measles vaccination Percentage of children aged 12 to 23 months who received measles +, #

Contraceptive prevalence (Modern method)

Percentage of currently married women aged 15 to 49 who currently use a modern method of contraception

Antenatal care visits- 4 or more (ANC4+)

Percentage of mothers aged 15 to 49 who had a live birth in the past 5 years who received at least 4 antenatal care visits from any skilled personnel during pregnancy for the most recent birth

Skilled birth attendant (SBA) Percentage of mothers aged 15 to 49 who had a live birth in the past years who were attended by skilled health attendant

Notes: * Based on WHO 2006 growth standards are used to calculate z-scores. + Immunizations are either verified by card or based on recall of respondent. # Data not collected in 2014 mini-DHS survey.

We selected six health status indicators and five health intervention indicators.

The five health status indicators include three child undernutrition and two mortality

indicators, namely stunting, wasting, underweight, neonatal mortality rate (NMR),

7

infant mortality rate (IMR), and under-five mortality (U5MR). On health interventions,

we included child and maternal health services. Two of the child health services

indicators included in this study are child immunization indicators: measles vaccination

and full immunization. The other three are maternal health services, namely prevalence

of modern contraceptive use by married women, four or more antenatal care visits from

a skilled professional (ANC4+), and delivery assistance from a skilled birth attendant

(SBA).

2.2 Data sources and variable construction

We used four demographic and health surveys (DHS) 2000, 2005, 2011 and 2014.5 The

surveys implemented in different periods follow standard and comparable

questionnaires. The data provide nationally representative information on the health

outcomes we selected for this study. The surveys also allow generating estimates for

rural and urban areas and regions. In addition, the timeline of the surveys in general

overlaps the larger part of the MDGs period.

The DHS data are organized in different files. Household level characteristics,

such as wealth and housing conditions (e.g., water and sanitation facilities), are from

the household data. Quintiles and other transformed variables, such as bottom 40% and

top 60% wealth categories, are computed from the household wealth index that is

5 The 2014 survey did not cover immunizations. The analysis for immunization indicators (full immunization and measles immunization) is based on the first three surveys.

8

available with the data. 6 The DHS wealth index is constructed from selected household

level variables, including housing characteristics, ownership of land and selected

durable goods (Rutstein and Johnson, 2004).

For child undernutrition analysis, we computed anthropometric indicators from

age, height/length and weight data following the WHO 2006 growth standards. We

calculated height-for-age, height-for-weight, and weight-for-age z-scores and then

stunting, wasting and underweight levels for children aged 0 to 59 months. Child

mortality rates (NMR, U5MR and IMR) are calculated using the standard DHS

methodology, using all data on all children deaths in the 5-year period preceding the

survey (Rutstein and Rojas, 2006). We calculated prevalence of modern contraception

use by currently married women, ANC4+, and SBA from the women’s (aged 15-49

years) questionnaire.

2.3 Data Analysis

We examined differential progress for groups by wealth ranking, place of residence and

education. For wealth groups, we looked at socioeconomic inequalities in health by

wealth ranking between the worse off (bottom 40%) and the better off (top 60%) and

between the poorest (1st quintile) and the richest (5th quintile). The spatial dimension we

investigated is place of residence: rural vs. urban residents. On education, we considered

6 For DHS 2000, we constructed quintiles from the wealth index. The bottom 40% comprises the first and the second quintiles, i.e., the poorest and the next poorest quintiles, and the top 60% represents the remaining three quintiles. The interest on the Bottom 40% of the wealth group is following the World Bank Gorup’s current goals to end extreme poverty and promote shared prosperity. http://www.worldbank.org/content/dam/Worldbank/document/WB-goals2013.pdf

9

mothers’ education on children’s nutrition status and mortality and own education on access to

health care.

For a more complete picture, we looked at a combination of approaches that are

often used in inequality studies because each approach has its own limitations and can

lead to different conclusions (e.g., Moser et al., 2007 and Mackinnon et al., 2014). We

computed absolute and relative inequalities from rate differences and rate ratios.

However, differences and ratios between different groups do not take into account

inequalities by the whole population. We, therefore, examined inequalities using

concentration indexes (CI) (Kakwani et al., 2002; O’Donnell et al., 2008). We used

concentration curves to illustrate the movement of socioeconomic inequalities in health

between the earlier and latest surveys.7 Following the procedure in O’Donnell (2006)

and O’Donnell et al. (2008), we conducted tests of dominance between concentration

curves of the outcomes for the earlier and latest surveys. The welfare measure we used

for CI analysis is the DHS wealth index. Our analysis also includes decomposition of

wealth-related inequalities in health.8

7 Earlier survey year refers to the 2000 DHS. The latest survey is the 2011 DHS for immunization indicators (full immunization and measles vaccination) and the 2014 DHS is considered as the latest survey for the rest of the indicators ( undernutrition and mortality indicators, contraception, ANC4+, and SBA). The concentration curve shows the plot of cumulative proportion of a health outcome against the proportion of the population ranked by a welfare measure. 8 We used STATA version 13.0 for most of the analyses, including trend analysis, rate difference and rate ratio analysis, concentration indexes, concentration curves and dominance tests. We used ADEPT Health to decompose the inequalities.

10

3. Results

3.1 Progress in MNCH

Over the last two decades, there has been considerable improvement in MNCH

outcomes, though there still exist high levels of child undernutrition and mortality rates

and limited coverage of maternal and health services. This holds true for all the health

status and health service indicators analyzed in this study. Table 2 presents the

progress at the national level and for rural and urban areas. The progress over the

period is shown by a decline in ill health (under nutrition and mortality) and an

increase in the coverage of health services (immunizations and maternal health

services). For example, in the 2014 DHS, about 48.2% (6.3 million) of children under

the age of five years were malnourished as measured by at least one of the three

standard indicators of undernutrition: stunting, underweight or wasting. However, the

current rates of undernutrition are much improved when compared to the 64.4% in the

2000 DHS. This decline, by 15 percentage points, is roughly equivalent to improvements

in the nutritional status of over 2 million children under the age of five years. 9

Looking at each indicator separately, in DHS 2000, at the national level, over half

of children under five years old were stunted and over 4 in 10 were underweight (Table

2). Over the years covered by the surveys, however, stunting declined by 17 percentage

points (from 57% to 40.6%); and prevalence of underweight declined by 15 percentage

points (from 41.9% to 26.6%). There has also been a significant reduction in wasting,

9 The estimate is based on the 97.1 million total population in 2013 (World Bank data) and the proportion of the age group which is 13.6 % (DHS 2014) and the current prevalence of undernutrition which is 48% (DHS 2014).

11

which declined by about 4 percentage points, from 12.5% in 2000 to 8.5% in 2014.

Another important measure of child health improvement over the last 15 years is the

declining trend in child mortality rates. During this period, U5MR declined by over 83

deaths per 1,000 live births, from 164 deaths in 2000 to 80 deaths in 2014, and IMR

declined by about 35 deaths per 1,000 live births, from 95.9 death per 1,000 live births in

2000 to 59.3 death per 1,000 live births in 2014. 10

Table 2. Trends in MNCH outcomes in Ethiopia, 2000-2014

National Rural Urban

Earlier Survey

Latest Survey Change

Earlier Survey


Earlier Survey


Stunting 57.0 40.6 -16.4 58.2 42.5 -15.7 46.5 27.0 -19.5

Wasting 12.5 8.9 -3.6 13.0 8.9 -4.1 7.8 9.0 1.2

Underweight 41.9 26.6 -15.3 43.4 27.8 -15.6 28.4 18.7 -9.7

NMR 48.4 34.3 -14.1 48.8 32.8 -16.0 44.5 44.4 -0.1

IMR 95.9 59.3 -36.6 114.2 65.7 -48.5 96.3 60.7 -35.6

U5MR 163.9 80.1 -83.8 191.3 91.8 -99.5 148.3 67.3 -81.0

Measles vaccination 27.0 56.7 29.7 43.2 48.5 5.3 63.1 79.6 16.5

Full immunization 14.6 24.9 10.3 11.2 20.9 9.7 43.2 48.5 5.3

Contraceptive 7.2 45.2 38.0 3.8 42.0 38.2 32.9 61.8 28.9

ANC4+ 10.4 24.2 13.8 3.2 37.1 33.9 28.5 56.2 27.7

SBA 5.7 15.5 9.8 2.3 9.1 6.8 34.6 58.4 23.8 Source: Authors’ compilation from the DHS (2000, 2005, 2011 & 2014) data. Note: The values are number of births per 1000 live births for NMR, IMR and U5MR and percentages for the rest. Earlier survey year is 2000 for all indicators. Latest survey year is 2014 for all but full immunization and measles vaccination, for which our latest source of information is the 2011 DHS.

10 The mortality results reported here are based on birth history records collected as part of DHSs. Birth histories usually go back 3-5 years before the date of the survey. Hence, the results might refer to a period 1.5- 2.5 years before the survey report date. It is more than likely that under- five mortality level in 2014 could be even lower. The main focus of the current paper is on the distribution, which is consistent over the four surveys. However, for the actual levels of mortality readers can also refer to the UN Inter Agency Group estimate http://www.unicef.org/publications/files/Child_Mortality_Report_2015_Web_8_Sept_15.pdf accessed on 7 November 7, 2015 (UNICEF, 2015).

12

Table 2 also presents progress in child immunization and maternal health services.

For example, full immunization coverage increased by over 10 percentage points, from

14.6% in 2000 to 24.9% in 2011. During the same period, measles vaccination coverage

increased by 30 percentage points, from 27% in 2000 to 57% in 2011. The relatively

better performance in measles vaccination might be because of additional deliveries

through outreach campaigns that tend to reach more people (Barros et al., 2012). There

has also been some expansion in coverage of all three maternal health services. For

example, modern contraceptive use by married women (age 15-49 years) was just 6% in

2000 and increased to 40% in 2014. During the same period, antenatal care visits (4 or

more) increased by 14 percentage points, from 10% in 2000 to 24% in DHS 2014.

Although still very low in terms of its coverage, the skilled birth attendant rate also

increased by over 10 percentage points from 5% in 2000 to about 16% in DHS 2014.

3.2 Inequalities

One important observation from the overall trends in MNCH in Ethiopia is that services

expanded and health outcomes improved broadly. However, our objective is on the

distribution of the progress. How have different groups benefited from this progress?

To answer this question, we disaggregated the trends for different groups, using

different approaches of inequality analyses. We looked at the inequalities by wealth

groups, place of residence and education. We analyzed the trends of overall progress in

13

MNCH (section 3.2.1). We also looked at the trends in rate differences (section 3.2.2),

rate ratios (Section 3.2.3), and concentration indexes and curves (section 3.2.4). 11

3.2.1 Trends in MNCH for different groups

We disaggregated the trends in MNCH outcomes by selected background

characteristics: rich vs. poor or top 60% compared to bottom 40%, wealth quintiles,

place of residence (urban and rural) and mother’s education. All the curves in each

graph plot the values for all survey years and the 95% confidence intervals. In each

graph, the vertical lines perpendicular to the years in the horizontal axis are 95%

confidence intervals. Figures 1a through 1d present the results.

Figure 1a is on nutrition; Figure 1b - child mortality; Figure 1c- immunization;

and Figure 1d is on modern contraceptive use by married women, ANC4+, and SBA.

In all cases, outcome curves are downward and health services curves are upward

sloping. This shows improvement in MNCH in all indicators on average. However, the

distance between the curves and the slope of each curve in each graph shows that the

performance has been different for different outcomes. For example, in Figure 1a, while

the reduction in undernutrition occurred for both the bottom 40% and the top 60%, the

curve for the top 60% is steeper than the bottom 40%. Similarly, the curve for the top

20% is steeper. Some four observations are noted from the disaggregated graphs in

undernutrition rates over the period covered by the surveys. One is that the income-

based inequality is starker because the poor–rich differences are driven by the richest 11 Mother’s education is used in child health indicators, including stunting, wasting, underweight, IMR , UFMR, measles vaccination and full immunization, while own education is used in maternal health services, including contraception,ANC4+, and SBA.

14

quintile. Proportionally more reductions in undernutrition occurred among children

from the richest quintile. The rest, from the 1st to the 4th quintile, are within the 95%

confidence intervals. Second, the widening income-related inequality in health is

stronger in the long- than short-term nutritional status because the absolute gap clearly

widened for stunting but not for wasting and underweight. Third, the rural–urban gaps

are either constant or narrowing. Fourth, education followed the pattern exhibited in

income; the improvement was higher among children whose mothers have more

education, secondary education and above.

Figure 1b shows the trends for child mortality (NMR, IMR and U5MR). Child

morality is an important outcome that the country has substantially reduced over the

0

20

40

60

80

100

2000 2005 20112014

rich (top 60%)poor (bottom 40%)

Rich/Poor

0

20

40

60

80

100

2000 2005 20112014

richestrichermiddlepoorpoorest

Wealth Quintile

0

20

40

60

80

100

2000 2005 20112014

urbanrural

Urban/Rural

0

20

40

60

80

100

2000 2005 20112014

secondary+primaryno education

Mother's Education

Stunting

0

20

40

60

80

100

2000 2005 20112014


Rich/Poor

0

20

40

60

80

100

2000 2005 20112014


Wealth Quintile

0

20

40

60

80

100

2000 2005 20112014

urbanrural

Urban/Rural

0

20

40

60

80

100

2000 2005 20112014


Mother's Education

Wasting

0

20

40

60

80

100

2000 2005 20112014


Rich/Poor

0

20

40

60

80

100

2000 2005 20112014


Wealth Quintile

0

20

40

60

80

100

2000 2005 20112014

urbanrural

Urban/Rural

0

20

40

60

80

100

2000 2005 20112014


Mother's Education

Underweight

Note: Computed from DHS data. Income is based on wealth index. Vertical lines in the graphs are 95% confidence intervals.

(2000-2014)

Figure 1a. Trends in child malnutrition by wealth ranking, residence and mother's education

15

last two decades. The graphs show the reductions in mortality on average. The

reduction is the smallest in NMR; the curves are less steep when compared to the

curves in IMR and U5MR. Figure 1b also shows that the reductions by income group

first exhibited reversals from a pro-poor trend (2000-2005) to a pro-rich inequality trend

(2005-2011) and then back to pro-poor inequality trend (2011 to 2014). Overall, the trend

presented in Figure 1b does not show a clear distinction in the distribution of

reductions in mortality by income group.

In Figure 1c, the trends in immunization services show that the expansion has

been for all groups by income, place of residence and mother’s education. The curves

0

40

80

120

160

200

2000 2005 2011 2014

top60%bottom40%

Rich/Poor

0

40

80

120

160

200

2000 2005 2011 2014

poorestpoormiddlericherrichest

Wealth Quintile

0

40

80

120

160

200

2000 2005 2011 2014

urbanrural

Urban/Rural

0

40

80

120

160

200

2000 2005 2011 2014

no educationprimarysecondary+

Mother's Education

NMR

0

40

80

120

160

200

2000 2005 2011 2014

top60%bottom40%

Rich/Poor

0

40

80

120

160

200

2000 2005 2011 2014


Wealth Quintile

0

40

80

120

160

200

2000 2005 2011 2014

urbanrural

Urban/Rural

0

40

80

120

160

200

2000 2005 2011 2014


Mother's Education

IMR

0

40

80

120

160

200

2000 2005 2011 2014

top60%bottom40%

Rich/Poor

0

40

80

120

160

200

2000 2005 2011 2014


Wealth Quintile

0

40

80

120

160

200

2000 2005 2011 2014

urbanrural

Urban/Rural

0

40

80

120

160

200

2000 2005 2011 2014


Mother's Education

U5MR

Note: Combiled from DHS data

(2000-2014)Figure 1b.Trends in child mortality by wealth ranking, residence and mother's education

16

are upward sloping, showing expansion in immunization coverage over the period. The

slope of the curves also shows that the rate of expansion was more rapid in the latest

years than in earlier years. The figure also shows that expansion of the immunization

coverage has been proportionally higher for the richest quintile.

Figure 1d presents trends for maternal health services. It shows that utilization of

modern contraception expanded for all groups. However, expansions in the utilization

of antenatal care visits and delivery assistance for a skilled professional have been

largely for the better off, for women in urban areas and for women with some

secondary education and above. The relatively better performance in contraceptive use

0

20

40

60

80

100

2000 2005 20112014

top 60%bottom 40%

Rich/Poor

0

20

40

60

80

100

2000 2005 20112014


Wealth Quintile

0

20

40

60

80

100

2000 2005 20112014

urbanrural

Urban/Rural

0

20

40

60

80

100

2000 2005 20112014


Mother's Education

Full immunization

0

20

40

60

80

100

2000 2005 20112014

top 60%bottom 40%

Rich/Poor

0

20

40

60

80

100

2000 2005 20112014


Wealth Quintile

0

20

40

60

80

100

2000 2005 20112014

urbanrural

Urban/Rural

0

20

40

60

80

100

2000 2005 20112014


Mother's Education

Measles vaccination

Note: Computed from DHS data. Income is based on wealth index. Vertical lines in the graphs are 95% confidence intervals.

(2000-2011)

Figure 1c. Trends in immunization services by wealth ranking, residence and mother's education

17

among rural women might be due to health extension workers in rural areas whose

major activities include delivering family planning services.

3.2.2 Absolute inequalities: Rate differences

Further to the graphical illustrations presented in the previous section, we examined

whether the changes in absolute gaps between different groups were significant. We

tested if these differences between groups at the earlier survey were significantly

different to that of the differences at the latest survey available to the indicator. Tables

3a & b present the results. The values in the tables are percentage points except for

NMR, IMR and U5MR where the number of births per 1,000 live births is reported. For

all indicators, the earlier survey report year is 2000 (the first DHS). The latest survey

report year is 2014 (the mini DHS) for the nutrition and mortality indicators and for the

0

20

40

60

80

100

2000 2005 20112014


Rich/Poor

0

20

40

60

80

100

2000 2005 20112014


Wealth Quintile

0

20

40

60

80

100

2000 2005 20112014

urbanrural

Urban/Rural

0

20

40

60

80

100

2000 2005 20112014


Education

Contraception

0

20

40

60

80

100

2000 2005 20112014

rich (top 60%)

poor (bottom 40%)

Rich/Poor

0

20

40

60

80

100

2000 2005 20112014


Wealth Quintile

0

20

40

60

80

100

2000 2005 20112014

urbanrural

Urban/Rural

0

20

40

60

80

100

2000 2005 20112014

secondary+primary

no education

Education

ANC4+

0

20

40

60

80

100

2000 2005 20112014


Rich/Poor

0

20

40

60

80

100

2000 2005 20112014


Wealth Quintile

0

20

40

60

80

100

2000 2005 20112014

urbanrural

Urban/Rural

0

20

40

60

80

100

2000 2005 20112014


Education

SBA

Note: Computed from DHS data. Income is based on the household's wealth index. The vertical lines in the graphs are 95% confidence intervals.

(2000-2014)

Figure 1d. Trends in maternal health services by wealth ranking, residence and education

18

three maternal health services included in this study, namely contraception, ANC4+,

and SBA. As indicated earlier, the mini-DHS did not cover immunization indicators.

Therefore, the latest survey for these indicators is the 2011 DHS.

Table 3a. Trends in absolute gaps MNCH between poor and rich households, 2000-2014

Diff: Bottom40%-Top60% Diff: Poorest-Richest Earlier Survey

Latest Survey P-value

Earlier Survey


Stunting 4 10 0.050 12 20 0.123 Wasting 0 3 0.994 3 4 0.914 Underweight 5 8 0.049 13 17 0.574 NMR -12.9 -13.2 NA -8.6 -8.5 NA IMR -24 -3 NA -10 -1 NA U5MR -19 3 NA -2 11 NA Measles Vaccination -16 -12 0.024 -34 -33 0.016 Full Immunization -11 -12 0.290 -28 -33 0.973 Contraceptive -7 -14 0.096 -23 -26 0.136 ANC4+ -10 -24 0.488 -31 -52 0.770 SBA -8 -19 0.130 -25 -51 0.632

Authors’ compilation from the DHS (2000, 2005, 2011 & 2014) data. Note: The p-values are for the F-test of the difference-in-differences (Diff-in-Diff). The first part of the results is for the test: Is [(bottom40% in the Latest Survey) - (top600% in the Latest Survey)] equal to [(bottom40% in the Earlier Survey) - (top60% in the Earlier Survey)]? And, in the second part the results are based on the following test: Is [(bottom20% in the Latest Survey) - (top20% in the Latest Survey)] equal to [(bottom20% in the Earlier Survey) - (top20% in the Earlier Survey)]? Test is not applicable for NMR, IMR and U5MR.

Table 3a shows the gap between the poor and the rich, and between the poorest

and the richest.12 Therefore, a positive value for an ill health outcome (child

undernutrition or child mortality) shows a pro-rich absolute inequality, i.e., child

undernutrition rates and mortality levels were lower among children from the better off

households. Likewise, a negative value in any of the immunization and maternal health

12 The poor and the rich classification refers to the bottom 40% and the top 60% respectively; whereas, the poorest and the richest classification refers to the bottom 20% (the first wealth quintile) and the top 20% (the fifth quintiles).

19

service indicators (good health outcomes) implies a pro-rich inequality. These

socioeconomic differences are expected. However, the trends in the differences are

relevant for interventions that could be adopted. Therefore, our interest is to show if the

absolute gaps were widening or narrowing.

The results, in Table 3a, show that there has been a widening absolute pro-rich

inequality between the poor and the rich in child nutritional outcomes. The pro-rich

inequality was observed in all the three child nutrition status indicators. It is significant

in stunting and underweight cases. In DHS 2000, the level of child stunting for the poor

was higher by about 4.2 percentage points. In DHS 2014, the difference between the

poor and the rich was 10 percentage points. Similarly, the gap in underweight increased

from 5 to 8 percentage points. In both cases (stunting and underweight), the differences

at the earlier survey and the latest survey were significantly different [p<0.05], implying

a widening pro-rich inequality. Similarly, the declines in child mortality have been

sharper /faster for the rich. For both IMR and U5MR, the differences at the earlier

survey were actually more pro-poor and they crossed the line of equality in favor of the

better off.

On the services side, there has been a larger increase in modern contraceptive use

by married women among the better off. On all others, however, either the

improvement has been pro-poor (e.g., measles vaccination [p<0.5]) or no indication of

significant widening in inequality. Given the overall improvement in the mean, this

suggests that both the poor and the rich benefited from the expansions in MNCH

services.

20

Other important socioeconomic determinants of MNCH outcomes and services

are education and place of residence. We examined differences in health inequalities by

these two outcomes (Table 3b). An important observation in Table 3b is that of the

improvement in the rural-urban gap. In all the indicators, the rural–urban difference in

health inequality has either improved (narrowing inequalities) or shown no significant

indication of worsening or widening inequalities. Although it is not as strong as the

spatial dimension, there is also a similar profile by education-based health inequalities.

The results for most of the indicators show that education-related absolute inequalities

in health have been narrowing. The exceptions are undernutrition indicators, where

mixed results were found.

Table 3b. Trends in absolute gaps MNCH inequalities by education and place of residence, 2000-201

Diff: Primary & below- Secondary+ Diff: Rural-Urban Earlier Survey


Earlier Survey


Stunting 22 23 0.649 12 16 0.147 Wasting 6 2 0.439 5 0 0.347 Underweight 18 22 0.022 15 9 0.012 NMR 21.4 17.6 NA 4.3 -11.6 NA IMR 31 27 NA 10 -12 NA U5MR 75 55 NA 38 5 NA Measles Vaccination -37 -35 0.112 -40 -27 0.105 Full Immunization -34 -34 0.506 -32 -28 0.588 Contraceptive -32 -23 0.362 -29 -20 0.110 ANC4+ -42 -49 0.411 -38 -47 0.545 SBA -41 -64 0.131 -32 -49 0.376

Authors’ compilation from the DHS (2000, 2005, 2011 & 2014) data. Note: The p-values are for the F-test of the difference-in-differences (Diff-in-Diff). Tests for education and rural-urban difference in differences are: Is [(less educated (primary and none)in the Latest Survey)- educated (secondary+) in the Latest Survey)] equal to [((less educated (primary and none) in the Earlier Survey)-( educated (secondary+) in the Earlier Survey)]? And, in the second panel of Table 4: Is [(rural in the Latest Survey) - (urban in the Latest Survey)] equal to [(rural in the Earlier Survey) - (urban in the Earlier Survey)]? Test is not applicable for NMR, IMR and U5MR.

21

3.2.3 Relative inequalities: Rate ratios

Another perspective for analyzing health inequalities is using the rate ratio (RR). RR

measures relative gap, i.e., if one group is more or less likely to have (or receive) a

certain outcome (or service) than the other group. An RR value greater than one

indicates a pro-rich inequality for an outcome. It shows a pro-poor inequality if the

outcome is defined positively, such as immunization and maternal health services

coverage. The correlations between these health indicators and the determinants, such

as income, education and location, are well-established. The rich are more likely to have

a better outcome compared with the poor. Similarly, urban residents are more likely to

utilize maternal and child health services than rural residents. The interest, however, is

in the changes in these RRs over time. Similar to the absolute gaps discussed earlier, the

trends in these ratios reveal important information for policy and interventions. Tables

4a & b present the results. The p-values in the tables are based on the test of the equality

of the RRs.

All the undernutrition indicators are greater than unity in both cases, implying

that children from the poor households (bottom 40% or bottom 20%) have a higher rate

of being undernourished compared to children from rich households (top 60% or top

20%). Similarly, children and women from the poor households are less likely to receive

the health services analyzed. The poor face lower odds of receiving immunizations,

using modern contraceptives, making required antenatal visits, and having deliveries

attended by a skilled health professional. Although these differences might be expected,

the changes in the relative inequalities over time are relevant to policy because it

22

indicates whether the poor are catching up or being left behind. For example, in DHS

2000, children from the bottom 40% had an almost equal chance of being stunted,

wasted and underweight (Table 4a). However, in DHS 2014, the chance of being

undernourished for this group is 27 to 40 percent higher than those of children from the

top 60%.

Table 4a.Trends in relative gaps MNCH between poor and rich households, 2000-2014

Ratio: Bottom40%/Top60% Ratio: Poorest/Richest Earlier Survey


Earlier Survey


Stunting 1.08 1.27 0.077 1.24 1.71 0.096 Wasting 1.03 1.40 0.991 1.37 1.58 0.861 Underweight 1.12 1.34 0.090 1.41 2.02 0.475 NMR IMR 0.78 1.31 NA 0.88 1.34 NA U5MR 0.99 1.41 NA 0.99 1.52 NA Measles Vaccination 0.51 0.80 0.038 0.35 0.59 0.008 Full Immunization 0.42 0.60 0.397 0.21 0.34 0.825 Contraceptive 0.32 0.72 0.425 0.12 0.54 0.028 ANC4+ 0.29 0.30 0.158 0.12 0.14 0.257 SBA 0.13 0.21 0.026 0.03 0.08 0.037

Source: Authors’ compilation from the DHS (2000, 2005, 2011 & 2014) data. Note: The p-values are for the F-test of the equality of rate ratios. The test in the first part of the table: Is [(bottom 40%, in the Latest Survey) / (top 60%, in the Latest Survey)] equal to [(bottom 40%, in the Earlier Survey) / (top 60%, in the Earlier Survey)] ? Similarly, in the second part of the table the quality of RR tests are: Is [(bottom 20%, in the Latest Survey) / (top 20%, in the Latest Survey)] equal to [(bottom 20%, in the Earlier Survey) / (top 20%, in the Earlier Survey)] ? Test is not applicable for IMR and U5MR. On the services side, over the study period, relative health inequalities declined

for coverage of immunization services. For example, for measles vaccination, the RR in

the latest survey is significantly different to that of the RR in the earlier survey. The RR

changed from 0.51 to 0.8 [p <0.05] in favor of the bottom 40% and from 0.35 to 0.59

[p<0.01] in favor of the bottom 20% (Table 4a). Other services where wealth-related

23

inequalities significantly declined (or a significant increase in RRs) include modern

contraceptive use and SBA. The poorest to the richest RRs of modern contraceptive use

by married women increased from 0.12 to 0.54 [p<0.05]. For SBA, both the poor to the

rich RR and the poorest to the richest RRs increased significantly, respectively from 0.13

to 0.21 [p<0.05] and 0.03 to 0.08 [p<0.05]. However, the improvement in relative gap by

wealth group is not significant for ANC visits.

Table 4b. Trends in relative gaps MNCH inequalities by education and place of residence, 2000-2014

Ratio Primary & below/Secondary+ Ratio: Rural/Urban Earlier Survey


Earlier Survey


Stunting 1.60 2.30 0.519 1.25 1.58 0.102 Wasting 1.80 1.29 0.279 1.68 0.98 0.432 Underweight 1.75 4.71 0.001 1.53 1.48 0.036 NMR 1.76 2.00 NA 1.10 0.74 NA IMR 1.46 1.69 NA 1.19 1.27 NA U5MR 1.81 2.27 NA 1.29 1.07 NA Measles Vaccination 0.41 0.61 0.012 0.36 0.66 0.008 Full Immunization 0.27 0.40 0.363 0.26 0.43 0.822 Contraceptive 0.15 0.65 0.137 0.11 0.68 0.036 ANC4+ 0.16 0.30 0.225 0.14 0.27 0.118 SBA 0.08 0.16 0.290 0.07 0.16 0.029

Source: Authors’ compilation from the DHS (2000, 2005, 2011 & 2014) data. Note: The p-values are for the F-test of the equality of rate ratios. The first part of the table is based on: Is [(primary and none, in the latest survey) / (secondary and above, in the latest survey)] equal to [(primary and none in the Earlier Survey) / (secondary and above, in the Earlier Survey)]? Similarly, in the second part of the table the quality of RR tests are: Is [(rural, in the Latest Survey) / (urban, in the Latest Survey)] equal to [(rural, in the Earlier Survey) / (urban, in the Earlier Survey)]? Test is not applicable for NMR, IMR and U5MR

Table 4b presents the education and spatial dimensions of relative gaps in health.

The results, in general, indicate that there has been narrowing relative gap by education

and place of residence for all services and nutrition and mortality indicators. Nine out

of ten indicators show a narrowing relative gap between rural and urban areas. The

24

improvement is significant for most of the indicators. One indicator that significantly

moved to the opposite direction is stunting.

3.2.4 Concentration indexes and concentration curves

In this section, we provide a more comprehensive analysis of wealth- related

inequalities in MNCH. We present results from concentration indexes (Tables 5a & 5b).

The Tables provide information on the survey-to-survey progression. Negative and

positive values are due to the definition of the health outcome under consideration. We

focus on the magnitudes in absolute terms. A smaller value in absolute terms indicates

improvement in health equality. Results in each outcome show no differences in the

standard errors across the survey years. This facilitates comparisons of wealth- related

inequalities over time (Koolman and van Doorslaer, 2004).

25

Table 5a. Concentration indexes of selected child health status indicators in Ethiopia (2000-2014)

Year CI SE norm CI SE abs CI SE Stunting

2000 -0.033 0.007 -0.076 0.016 -0.019 0.004 2005 -0.042 0.011 -0.083 0.022 -0.021 0.005 2011 -0.064 0.009 -0.115 0.016 -0.028 0.004 2014 -0.085 0.015 -0.144 0.025 -0.035 0.006

Wasting 2000 -0.034 0.020 -0.039 0.022 -0.004 0.002 2005 -0.117 0.028 -0.133 0.032 -0.015 0.004 2011 -0.159 0.022 -0.176 0.025 -0.016 0.002 2014 -0.099 0.037 -0.109 0.041 -0.009 0.003

Underweight 2000 -0.054 0.009 -0.093 0.016 -0.023 0.004 2005 -0.075 0.015 -0.113 0.022 -0.025 0.005 2011 -0.129 0.012 -0.182 0.017 -0.038 0.003 2014 -0.118 0.019 -0.161 0.027 -0.032 0.005

NMR 2000 0.050 0.052 0.002 2005 0.036 0.037 0.001 2011 -0.085 -0.088 -0.003 2014 0.080 0.083 0.003

IMR 2000 0.038 0.042 0.004 2005 -0.016 -0.018 -0.001 2011 -0.083 -0.088 -0.005 2014 0.013 0.013 0.001

U5MR 2000 0.015 0.018 0.002 2005 -0.049 -0.056 -0.006 2011 -0.105 -0.114 -0.009 2014 -0.021 -0.023 -0.002

Source: Authors’ compilation from the DHS (2000, 2005, 2011 & 2014) data. Note: CI is concentration index; norm CI is normalized concentration index; and abs CI is absolute concentration index.

26

Table 5b. Concentration indexes of selected MNCH services in Ethiopia (2000-2014) Year CI SE norm CI SE abs CI SE Measles Vaccination 2000 0.242 0.029 0.331 0.040 0.065 0.008 2005 0.142 0.023 0.224 0.036 0.052 0.008 2011 0.096 0.016 0.223 0.036 0.055 0.009 Full Immunization 2000 0.345 0.044 0.403 0.051 0.050 0.006 2005 0.173 0.035 0.220 0.044 0.037 0.008 2011 0.223 0.031 0.297 0.042 0.056 0.008 ANC 2000 0.474 0.030 0.529 0.034 0.049 0.003 2005 0.467 0.025 0.530 0.028 0.056 0.003 2011 0.447 0.022 0.532 0.026 0.071 0.004 2014 0.382 0.024 0.503 0.032 0.092 0.006 Contraception 2000 0.496 0.039 0.534 0.042 0.036 0.003 2005 0.382 0.019 0.453 0.023 0.060 0.003 2011 0.243 0.014 0.347 0.019 0.073 0.004 2014 0.120 0.014 0.218 0.026 0.054 0.006 SBA 2000 0.665 0.037 0.705 0.040 0.038 0.002 2005 0.698 0.034 0.740 0.036 0.040 0.002 2011 0.650 0.028 0.722 0.031 0.065 0.003 2014 0.526 0.030 0.622 0.035 0.081 0.005

Source: Authors’ compilation from the DHS (2000, 2005, 2011 & 2014) data. Note: CI is concentration index; norm CI is normalized concentration index; and abs CI is absolute concentration index.

The CIs in all the three undernutrition indicators show widening pro-rich

inequality; as data from the DHS 2000-2014 show, the CI in absolute value more than

doubled for all undernutrition indicators. This is also illustrated in Figure 2a, which

compares concentration curves in two periods. The concentration curves for the latest

survey are on the top of the concentration curves for the earlier survey showing the

increase of wealth-related inequalities in favor of the better off. The test of dominance

between concentration curves in Figure 2a shows that the upward shift is significant in

27

the case of wasting and underweight, but not in stunting. However, the dominance test is

significant only by multiple comparison approach (mca) rule. No significant dominance was

found using the more strict, intersection union principle (iup) rule. Overall, the results

presented here are in line with the findings from absolute and relative inequality

analyses presented in earlier sections; the poor did not benefit equally from the

improvements in child nutritional status. On the other hand, in Figure 2b, the

concentration curves overlapped and in some cases crossed the line of equality.

0.2

.4.6

.81

cum

ulat

ive p

ropor

tion:

stu

ntin

g

0 .2 .4 .6 .8 1cumulative proportion: children under 5

Earlier Survey Latest Survey

Line of Equality

Stunting

0.2

.4.6

.81

cum

ulat

ive p

ropor

tion:

was

ting



Line of Equality

Wasting0

.2.4

.6.8

1cu

mul

ativ

e p

ropor

tion:

und

erw

eig

ht


Earlier Survey Latest SurveyLine of Equality

Underweight

Note: Computed from DHS 2000 (Earlier Survey) and mini-DHS 2014 (Latest Survey). Values in the horizontal axis are ranked by wealth

(2000-2014)Figure 2a. Concentration curves of selected child undernutrition indicators

28

Figure 2c illustrates the movement of wealth-related inequalities in maternal

health services and child immunizations. The curves, in both the earlier and latest

surveys, are below the line of equality indicating the presence of pro-rich inequality in

using these services. However, for all the MNCH services, the movement of the indexes

and the curves illustrate progression towards the line of equality. For a given survey

year, the CIs are smaller compared with the values in an earlier survey (Table 5).

However, the magnitude of the change is different for different services. The move

towards the line of equality is the biggest in modern contraception use; the standard CI

declined from 0.496 in DHS 2000 to 0.12 in 2014. A similar performance is also observed

in measles vaccination: standard CI declined from 0.242 in DHS 2000 to 0.096 in 2014.

0.2

.4.6

.81

cum

ulativ

e p

roport

ion: neo

nat

al d

eath

s

0 .2 .4 .6 .8 1cumulative proportion: births


Line of Equality

NMR

0.2

.4.6

.81

cum

ula

tive p

roport

ion:

infa

nt d

eat

hs



Line of Equality

IMR

0.2

.4.6

.81

cum

ula

tive p

ropo

rtio

n: u

nde

r-fiv

e d

eat

hs



Line of Equality

U5MR

Note: Computed from DHS 2000 (Earlier Survey) & DHS 2014 (Latest Survey). Values in the horizontal axis are ranked by wealth.

(2000-2014)Figure 2b. Concentration curves of NMR, IMR and U5MR

29

Despite the progress, the inequalities are still substantial in the case of ANC4+and SBA,

although, in both cases, the curves moved closer to the line of equality.

The test of dominance between concentration curves finds significant

improvements for measles vaccination, contraception, ANC4+, and SBA. The shift was

not significant for full immunization. In addition, only the improvement in SBA is

found to be significant when the stricter rule, iup, is used.

0.2

.4.6

.81

cum

ulat

ive

pro

port

ion

: un

derw

eigh

t



Line of Equality

Contraception

0.2

.4.6

.81

cum

ulat

ive

pro

port

ion

: rec

eive

d A

NC

0 .2 .4 .6 .8 1cumulative proportion: last birth in previous 5 years


Line of Equality

ANC4+

0.2

.4.6

.81

cum

. pr

opro

port

ion:

att

end

ed b

y sk

illed

per

sonn

el

0 .2 .4 .6 .8 1cumulative proportion: births in last 5 years


Line of Equality

SBA

0.2

.4.6

.81

cum

ula

tive

pro

port

ion:

with

me

asle

s va

ccin

atio

n

0 .2 .4 .6 .8 1cumulative proportion: children 12-23 months


Line of Equality

Measles Vaccination

0.2

.4.6

.81

cum

ula

tive

pro

port

ion:

with

full

vacc

inat

ion

0 .2 .4 .6 .8 1cumulative proportion: children 12-23 months


Line of Equality

Full Immunization

Note: Computed from DHS 2000 (Earlier Survey), DHS 2011 (Latest Survey for immunizations) and mini-DHS (2014-Latest Survey for Contraception, ANC4+ and SBA).Values in the horizontal axis are ranked by wealth

(2000-2014)

Figure 2c. Concentration curves of selected MNCH services

30

3.3 Decomposition of the inequalities

The analyses in earlier sections show that, despite improvements, there are still

considerable inequalities in some MNCH outcomes and services. In this section, we

examine the contributors of the wealth -related inequalities in most of the indicators.

Wealth- related inequality is less of a concern for mortality indicators (IMR and U5MR);

both show concentration indexes that are close to zero (Table 6a). What contributes to

the inequality in child nutritional status, immunization coverage, modern contraceptive

use, antenatal care visits and skilled birth attendant?13 What is the role of spatial,

household and individual characteristics in explaining health inequalities in Ethiopia?

What is the extent of inequity or unjustified inequality in MNCH outcomes? To answer

these questions, we decomposed the concentration indexes of each indicator using the

latest survey data, the Ethiopia mini-DHS 2014.

Tables 6a & 6b present the decomposition of concentration indexes. The

decompositions are standardized for demographic variables to compute the inequity.

Stunting, wasting, and underweight are standardized by age of the child, gender, and

birth order. Measles and full immunizations are standardized by age of the mother.

Health services, including contraceptive use, ANC 4+, and SBA are standardized by

age. Wealth quintiles, place of residence (urban=1) and region dummies, education,

31

and household size indicators are included in all models. We added improved water

and sanitation facilities in the child undernutrition models. 14

The three undernutrition indicators have about the same level of inequality.

However, the contributions of the various determinants are different in different

indicators. For example, wealth ranking explains most of the inequity in stunting and

underweight. Mother’s education, and the head of household’s education are the most

important contributors to the inequity associated with wasting. It is also worth noting

that that the contribution of wealth ranking is very small in wasting and underweight.

Other important contributors are spatial variables, including place of residence and

region. Other factors, including family size and water and sanitation, contributed very

little to the total inequity associated with undernutrition.

The decomposition of the inequity in service utilization shows that the existing

inequity is highest in SBA (0.525) and in ANC 4+ (0.380), followed by full immunization

services. It is much lower in measles vaccination (0.096) and contraception use by

married women (0.123). The household’s wealth ranking is the most important factor in

the inequity in the utilization of services. The next is education, although it is less

important in equity in contraception use. The values range from 0.03 in contraception to

0.085 in skilled birth attendance.

14 Dummy variables included for Region and Region-1 (Tigray) is the reference region. Residual regression error is part of the concentration index that is not explained by the standardizing and control variables. Residual missing data is discrepancy between the CI and its decomposition due to the dropping of observations when estimating the model. Inequality is total concentration index. Inequity/Unjustified inequality is the unjustifiable inequality in the sense that it is not caused by the standardizing demographic variables.

32

Table 6a. Decomposition of concentration index for child undernutrition, non-linear models

Stunting Wasting Underweight Standardizing (demographic) variables Child age in months -0.003 0.004 -0.004 Child sex (Male=1, Female=0) 0.000 0.001 0.000 Birth order number 0.001 -0.003 -0.006 Subtotal -0.002 0.002 -0.009 Control variables Quintiles of wealth index -0.062 0.001 -0.038 Education/ Mother's education level -0.017 -0.032 -0.037 Education of HH Head -0.010 -0.037 -0.017 Improved water source for drinking 0.010 -0.022 -0.006 Improved toilet facilities -0.002 0.016 0.007 Number of household members 0.002 0.008 -0.004 Number of children 5 and under 0.001 0.007 -0.002 Area of residence (Urban=1, Rural=0) -0.016 0.015 -0.002 Region -0.004 -0.035 -0.022 Subtotal -0.098 -0.086 -0.118 Residual: regression error 0.008 -0.008 0.011 Residual: missing data 0.008 -0.007 -0.002 Inequality (total) -0.085 -0.099 -0.118 Inequity/Unjustified inequality -0.091 -0.094 -0.107

Note: ADEPT software (health outcome module) is used to produce the results. The decompositions are based on mini-DHS 2014 Data.

Another observation from the decomposition is the differential effect of spatial

variables. The effect of place of residence (being urban) is more important in services

than in health status outcomes. This might be expected because there are proportionally

more health facilities with skilled professionals in urban than rural areas. Also, for

equity in MNCH services, the rural-urban classification is more important than region.

33

Table 6b. Decomposition of concentration index for maternal health services, non-linear models

Measles Vaccination

Full

Immuni- zation

Contra- ception

ANC4+ SBA

Standardizing (demographic) variables

Respondent Age (mother’s age) 0.000 0.000 -0.003 0.001 0.001 Subtotal 0.000 0.000 -0.003 0.001 0.001 Control variables Quintiles of wealth index 0.057 0.175 0.061 0.215 0.268 Education 0.036 0.036 0.013 0.060 0.085

Number of household members 0.001 0.001 0.000

-0.001 -

0.002 Number of children 5 and under 0.002 0.002 0.014 0.004 0.018 Area of residence (Urban=1, Rural=0) 0.013 -0.004 0.041 0.069 0.072 Region 0.002 0.017 -0.002 0.041 0.027

Subtotal 0.106 0.226 0.127 0.387 0.468

Residual: regression error -0.010 -0.003 -0.004 -0.007 0.057 Residual: missing data 0.000 0.000 0.000 0.000 0.000 Inequality (total) 0.096 0.223 0.120 0.382 0.526 Inequity/Unjustified inequality 0.096 0.223 0.123 0.380 0.525

Note: ADEPT software (health utilization module) is used to produce the results. The decompositions are based on DHS 2011 data for Measles and Full Immunization and mini-DHS 2014 data for contraception, ANC4+ and SBA.

4. Discussion and conclusion

Ethiopia is making significant progress in MNCH. The trends over the period covered

by DHS 2000-2014 period show considerable declines in child undernutrition and

mortality levels, as well as increases in health service coverage, such as immunization,

contraceptive use, antenatal care visits and assistance from a skilled birth attendant.

Although the current levels of services are still low and the health risks are still high

compared to similar countries in Africa,15 and hence much more needs to be done to

15 Child undernutrition and mortality levels are among the highest in sub-Saharan Africa. Health services coverage is among the lowest. For example, in 2011, full immunization

34

expand the services and improve health status, the improvements so far in these

outcomes provoke a further look at the distributions by different groups.

The main objective of our study has, therefore, been exploring the inequalities.

We examined differential progress in health status and health services utilization by

socioeconomic background, including wealth ranking of households, education and

place of residence. We looked at a combination of approaches: rate differences, rate

ratios and concentration indexes. The inequality analyses results are summarized in

Table 7. The second and third columns, respectively, show results from rate differences

and rate ratios. The third column reports direction of concentration indexes based on

Tables 5a & 5b. Finally, the last column reports the test of dominance between the

concentration curves of the outcome under consideration. Caution is, however, needed

as analyses from these differences and ratios are sensitive to initial values. For the

concentration index, an improvement is a decline in absolute terms. Improvement from

the concentration curves is based on the test of dominance (see Supplemental Annex

Table S1).

coverage in Ethiopia was 25% compared with 90% in Rwanda, 83% in Burundi, 81% in Burkina Faso and Malawi, to mention but a few. Similarly, birth in a health facility in 2011 was 11% in Ethiopia compared with 31-92% in the rest of the 17 similar countries with comparable data.

35

Table 7. Summary of results of trends in income related MNCH inequalities, (2000-2014)

Rate Differences Rate Ratio

CI

Test of dominance: concentration curves

Bottom 40% vs. Top 60%

Poorest (q1) vs. Richest (q5)

Bottom 40% vs. Top 60%

Poorest (q1) vs. Richest (q5)

mca rule iup rule

Panel A: Health Status

Stunting Worsened NS Worsened Worsened Worsened NS NS

Wasting NS NS NS NS Worsened Worsened NS

Underweight Worsened NS Worsened NS Worsened Worsened NS

NMR* NA NA NA NA Improved NA NA

IMR* NA NA NA NA Improved NA NA

U5MR* NA NA NA NA Improved NA NA

Panel B: Health Services

Measles Vaccination

Improved Improved Improved Improved Improved Improved NS

Full Immunization

NS NS NS NS Improved NS NS

Contraceptive NS NS NS Improved Improved Improved NS

ANC4+ NS NS NS NS Improved Improved NS

SBA NS NS Improved Improved Improved Improved Improved

Note: Table summarizes results of different approaches presented in Tables 3 thru 5 and the test of dominance. The test of dominance is based on O’Donnell (2006) and O’Donnell et al. (2008). The number of evenly spaced quintile points is 19 (from 5% to 95%) and the significance level is 5%. The dominance test rules, mca and iup, respectively denote multiple comparison approach and intersection union principle. NA is test not applicable. NS is no significant change from the earlier survey. * The improvement for mortality indicators is a progression of the distribution from a more pro-poor towards the line of equality.

There are four key messages of this paper. First, broadly, socioeconomic

inequalities in maternal health services (Contraception, ANC4+, and SBA) narrowed

over time. The result is robust to changes in method of analysis and measures of

socioeconomic characteristics; the inequality narrowed by wealth group and also place

of residence. Expansions in measles vaccination have also been pro-poor and pro-rural.

36

We also note a similar trend for full immunizations, but not as strong as measles

vaccination. What this suggests is that those outcomes that are directly dependent on a

strong health system seem to be performing well, including the role of health extension

workers who have been linked with improving sector performance in rural areas.

Second, in contrast with health services, we find mixed results on child health

status (undernutrition and mortality). Although there has been a substantial reduction

in child undernutrition, there has also been an increase in the inequality between the

poor and the rich. Child nutritional outcomes are affected by many factors, including

food security, feeding practice, availability of monitoring services and supplements,

which was not captured by this study so as to explain the increasing inequality. It is also

the case that multisector determinants play a key role in nutrition and in mortality.

Having considered this limitation, pro-rich progress in health status coupled with pro-

poor progress in health services would be due to differences in the quality of services

accessed by different groups, as well as the role of other socio-economic factors beyond

the health sector. This predicament might also point to the differential progress in other

important determinants of these health outcomes. However, for child mortality,

wealth-related health inequalities progressed from a much pro-poor inequality to more

equality. Although the direction is towards the pro-rich inequality, the distribution has

been around the equality line. It is important to note that various DHSs surveys have

37

shown inconsistent gradient between child mortality and wealth quintiles, which could

be due to data errors or the construction of the wealth index itself.16

Third, the prevailing wealth-related inequalities in some services are not

narrowing. Despite the pro-poor progress, the gap between the poor and the rich is still

large; the utilization of these services is more concentrated among richer individuals.

This is particularly the case for antenatal care visits and skilled birth attendance. For

these services, the concentration curves for the latest survey are far below the equality

line. This shows that the poor still faces lower odds of making the required antenatal

care visits from a skilled health professional and delivering with the assistance of a

skilled birth attendant. However, given the narrowing trends of the inequality, the poor

would benefit proportionally more from further expansion of these services.

Fourth, decomposition of inequalities shows that differences in wealth ranking are the

most important contributor to the inequity. Other major contributors to health inequity include

education. It is either maternal education in child health models or own education in maternal

health services. The decomposition also shows that location variables, including regional

differences and place of residence, matter.

16 Observations from child mortality data compiled from DHS stat compiler http://beta.statcompiler.com/

38

Supplemental Annex:

Figure- S1a. Inequality decomposition of child undernutrition

Note: The decomposition chart excludes standardizing demographic variables, including age, sex and birth order of the child.

Figure-S1b. Inequality decomposition of MNCH services

Note: The decomposition chart excludes age (mother’s age for immunizations and own age for services), which is the standardizing variable used the decomposition.

‐0.14

‐0.12

‐0.1

‐0.08

‐0.06

‐0.04

‐0.02

0

0.02

0.04

0.06

Stunting Wasting Underweight

Wealth Ranking

Education: HH Head

Education: Mother's

Residence:Urban

Region

Toilet: Improved

Drinking Water Source: Improved

Number of children 5 and under

Number of household members

‐0.1

0

0.1

0.2

0.3

0.4

0.5

MeaslesVaccination

FullImmunization

Contraception ANC4+ SBA

Number of householdmembers

Number of children 5 andunder

Region

Residence: Urban

Education Level

Wealth Ranking

39

Table S1. Test of dominance between concentration curves

Data 1 Survey Year

Data 2 Survey Year

Rule mca iup

Stunting 2000 2014 non-dominance non-dominance Wasting 2000 2014 2014 dominates 2000 non-dominance Underweight 2000 2014 2014 dominates 2000 non-dominance Measles Vaccination 2000 2011 2014 dominates 2000 curves cross Full Immunization 2000 2011 non-dominance curves cross Contraception 2000 2014 2014 dominates 2000 non-dominance ANC4+ 2000 2014 2014 dominates 2000 non-dominance SBA 2000 2014 2014 dominates 2000 2014 dominates 2000

Note: Test of dominance for concentration curves Figures 2a-d based on O’Donnell (2006) and O’Donnell et al. (2008). Significance level is 5% and the number of evenly spaced quintiles points is 19 (from 5% to 95%). The test is not applicable to the mortality concentration curves we developed in this study. However, curves cross in all illustrations of mortality concentration curves.

References

Ásgeirsdóttir, T. L., and D. Ó. Ragnarsdóttir. 2013. “Determinants of relative and

absolute concentration indices: evidence from 26 European countries.”

International Journal for Equity in Health, 12, 53.

Cutler, D.M, A.S. Deaton and A. Lleras-Muney. 2006. “The determinants of mortality.” J

Econ Perspectives, 20: 97–120.

Federal Ministry of Health (FMOH) (2013), Annual Performance Report-2012/13. Addis

Ababa, Ethiopia: Federal Ministry of Health.

ICF International Inc. 2013. Trends in Demographic and Reproductive Health Indicators in

Ethiopia. Calverton, Maryland, USA: ICF International Inc.

Kakwani N., A. Wagstaff and E. van Doorslaer. 1997. “Socioeconomic inequalities in

health: measurement, computation and statistical inference.” J Econ, 77: 87–103.

Kruk, M. E., M. R. Prescott and S. Galea. 2008. “Equity of Skilled Birth Attendant

Utilization in Developing Countries: Financing and Policy Determinants.”

American Journal of Public Health, 98(1):142–147.

40

McKinnon, B., S. Harper, J.S. Kaufman and Y. Bergevin. 2014. “Socioeconomic

inequality in neonatal mortality in countries of low and middle income: a

multicountry analysis.” The Lancet Global Health, 2( 3 ): e165 - e173.

Mirkuzie, A. H. 2014. “Exploring inequities in skilled care at birth among migrant

population in a metropolitan city Addis Ababa, Ethiopia; a qualitative study.”

International Journal for Equity in Health, 13(1), 110.

Moser, K, C. Frost and D.A. Leon. 2007. “Comparing health inequalities across time and

place rate ratios and rate differences lead to different conclusions: analysis of

cross-sectional data from 22 countries 1991- 2001.” Int J Epidemiol..,36: 1285–1291.

O’Donnell O., E. van Doorslaer , A. Wagstaff A, and M. Lindelow. 2008. Analyzing

Health Equity Using Household Survey Data: A Guide to Techniques and their

Implementation. Washington, DC: World Bank.

Onarheim, K. H., S. Tessema, K. A. Johansson, K. T. Eide, O. F. Norheim, and I.

Miljeteig. 2012. “Prioritizing Child Health Interventions in Ethiopia: Modeling

Impact on Child Mortality, Life Expectancy and Inequality in Age at Death.”

PLoS ONE, 7(8), e41521.

Quentin, W., O. Abosede, J. Aka, P. Akweongo, K. Dinard, A. Ezeh and L. Sundmacher.

2014. “Inequalities in child mortality in ten major African cities.” BMC Medicine,

12, 95.

Rutstein, S. O. and K. Johnson. 2004. “The DHS Wealth Index.” DHS Comparative

Reports, No. 6. Calverton, Maryland: ORC Macro.

Rutstein, S. O. and G. Rojas. 2006. Guide to DHS Statistics. Demographic and Health

Surveys. Calverton, Maryland: ORC Macro.

UNICEF. 2015. Levels & Trends in Child Mortality: Estimates Developed by the UN Inter-

agency Group for Child Mortality Estimation. New York: United Nations Children’s

Fund.

Victora C.G., A.J. Barros, H. Axelson, Z.A. Bhutta, M. Chopra, G.V. França, K. Kerber,

B.R. Kirkwood, H. Newby, C. Ronsmans, J.T.Boerma. 2012. “How changes in

coverage affect equity in maternal and child health interventions in 35

41

Countdown to 2015 countries: an analysis of national surveys.” Lancet 380

(9848): 1149–56.

Victora C.G., J.P. Vaughan, F.C. Barros, A.C. Silva and E. Tomasi. 2000. “Explaining

trends in inequities: evidence from Brazilian child health studies.” Lancet

356(9235): 1093-98.

Wagstaff, A. M. Bilger, Z. Sajaia and M. Lokshin. 2011. Health Equity and Financial

Protection : Streamlined Analysis with ADePT Software. Washington D.C.: World

Bank. https://openknowledge.worldbank.org/handle/10986/2306 License: CC

BY 3.0 IGO.

Wagstaff A, E. van Doorslaer and N. Watanabe. 2001. “On decomposing the causes of

health sector inequalities, with an application to undernutrition inequalities in

Viet Nam.” Policy Research Working Paper No. 2714. Washington D.C.: World

Bank.

Wagstaff, A. 2002. “Poverty and health sector inequalities.” Bulletin of the World Health

Organization, 80 (2).

Wilunda, C., G. Putoto, F. Manenti, M. Castiglioni, G. Azzimonti, W. Edessa, and B.

Criel. 2013. “Measuring equity in utilization of emergency obstetric care at

Wolisso Hospital in Oromiya, Ethiopia: a cross sectional study.” International

Journal for Equity in Health, (12) 27.

WHO Multicentre Growth Reference Study Group. 2006. WHO Child Growth Standards:

Length/height-for-age, weight-for-age, weight-for-length, weight-for-height and body mass index-for-age:

Methods and development. Geneva: World Health Organization .

http://www.who.int/childgrowth/publications/en/.

World Bank. 2012. Health Equity and Financial Protection Datasheet - Ethiopia.

Washington, D.C.: World Bank.

Yesuf, E. A. and R. Calderon-Margalit. 2013. “Disparities in the use of antenatal care

service in Ethiopia over a period of fifteen years.” BMC Pregnancy and Childbirth,

(13) 131.

Maternal and Child Health Inequalities in Ethiopia

Documents

Transcript of Maternal and Child Health Inequalities in Ethiopia